Effects of an Angiotensin-Converting–Enzyme Inhibitor, Ramipril, on Cardiovascular Events in High-Risk Patients
2000 · paper · Clinical trials
A randomized trial of ramipril versus placebo in 9,297 high-risk patients without known heart failure or low ejection fraction. Ramipril reduced the combined rate of heart attack, stroke or cardiovascular death (14.0% versus 17.8%).
3 findings
Substance Abstract gives cardiac arrest as "0.8 percent vs. 1.3 percent; relative risk, 0.63; P=0.03", while Table 4 prints "0.62 (0.41–0.94)", z −2.28, P 0.02 and the results text says "37 vs. 59; relative risk, 0.62; P=0.02".
- p. 1 Abstract, Results
cardiac arrest (0.8 percent vs. 1.3 percent; relative risk, 0.63; P=0.03)
- p. 5 Table 4, row Cardiac arrest
0.62 (0.41–0.94)
- p. 5 Secondary and Other Outcomes, text
had a cardiac arrest (37 vs. 59; relative risk, 0.62; P=0.02)
- p. 1 Abstract, Results
cardiac arrest (0.8 percent vs. 1.3 per-
Judge's note: Read the abstract on the page 1 image and Table 4 and the results text on the page 5 image. The same comparison (37 vs 59 patients, 0.8% vs 1.3%, ramipril 10 mg vs placebo) is printed two ways. The paper describes one analysis (log-rank test, Cox model stratified for vitamin E) and no other analysis set is named for the abstract; the low-dose footnote in Table 3 concerns the primary outcome only. Table and text agree with each other, so the abstract is the odd one out. The crude ratio of proportions is 0.628, which may be where 0.63 came from, but the paper's reported estimate is 0.62 in two places. No erratum is among the files.
Second check: another model agreed
Report a problem with this finding
Substance Text says patients stopped treatment because of a clinical event in "8.9 percent vs. 6.6 percent" (placebo vs ramipril), while Table 2 prints Clinical events "309 (6.7)" for ramipril and "418 (9.0)" for placebo.
- p. 3 Compliance, text, right column
mary or secondary outcome (8.9 percent vs. 6.6 per-
- p. 3 Table 2, row Clinical events
309 (6.7)
- p. 3 Table 2, row Clinical events
418 (9.0)
- p. 3 Compliance, text
a primary or secondary outcome (8.9 percent vs. 6.6 percent)
Judge's note: Read the text and Table 2 on the page 3 image. The text cites Table 2 for these reasons. The table's counts give 8.99% and 6.65% with the group sizes 4652 and 4645, matching the table's 9.0 and 6.7. The other percentages in the same two sentences (7.3/1.8, 1.9/1.5, 3.9/2.3) match the table exactly, so truncation is not the text's general practice; 8.99 cannot round to 8.9. No different denominator is stated. The difference is 0.1 percentage point in each group, so the effect on a reader is small.
Second check: another model agreed
Report a problem with this finding
Substance Figure 2 prints placebo-group incidence "18.7" for Male sex and "14.4" for Female sex; with 1201 women and 3451 men in the placebo group (Table 1) these give about 818 events, not the 826 placebo events in Table 3.
- p. 6 Figure 2, rows Male sex / Female sex, column Incidence of Composite Outcome in Placebo Group
18.7
- p. 6 Figure 2, row Female sex
14.4
- p. 3 Table 1, Female sex, placebo group
1201 (25.8)
- p. 4 Table 3, primary outcome, placebo group
826 (17.8)
Only the authors can say what is right here.
Judge's note: Read Figure 2 on the page 6 image, Table 1 on page 3 and Table 3 on page 4. Placebo men = 4652 − 1201 = 3451. The printed rates give 645.3 + 172.9 = 818.3 events; even at the top of the rounding range (18.75% and 14.45%) the total is 820.6, below 826. The other pairs whose placebo sizes are in Table 1 reproduce 826 within rounding (diabetes 826.0, hypertension 826.9, microalbuminuria 826.9), so the column holds crude rates and the sex pair is the outlier. If 14.4 is right the male rate would be about 18.9%. Sex counts are complete (6817 + 2480 = 9297). Which of the two values is misprinted cannot be told from the files; the difference is about 0.2 percentage points. 18.7 is also the value on the Cardiovascular disease row.
Second check: another model agreed
Report a problem with this finding
2 minor (typos and the like; most checked by the judge only)
Typo Table 4, Hospitalization for unstable angina: "0.98 (0.87– 1.10)" has a space after the en dash, unlike the other intervals in the table.
- p. 5 Table 4, Hospitalization for unstable angina, relative risk column
0.98 (0.87– 1.10)
Judge's note: Opened the Table 4 crop and the page 5 image. The gap after the dash is visible in the image and is absent from the other rows (for example "0.85 (0.77–0.94)" and "0.88 (0.70–1.10)"). Purely cosmetic; the values are unaffected.
Second check: none (judge only)
Report a problem with this finding
Typo "heart failure in patients with no evidence of impairment of left ventricular systolic dysfunction" should read "systolic function" (or "no evidence of left ventricular systolic dysfunction").
- p. 7 Discussion, left column, last paragraph
heart failure in patients with no evidence of impair-
- p. 7 Discussion, left column, last paragraph
ment of left ventricular systolic dysfunction
- p. 7 Discussion, paragraph beginning 'We also observed'
heart failure in patients with no evidence of impairment of left ventricular systolic dysfunction
- p. 7 Discussion, paragraph beginning "We also observed a reduction in the incidence of heart failure"
no evidence of impair-ment of left ventricular systolic dysfunction
Judge's note: Read the paragraph beginning "We also observed" on the page 7 image; the wording is as quoted, split across a line break. "Impairment of ... dysfunction" doubles the negative. The discussion's opening uses the correct form, "without evidence of left ventricular systolic dysfunction" (page 6). The meaning is recoverable from context.
Second check: none (judge only)